It is now apparent that regulation of blood vessel growth contributes to the classical actions of hormones on development, growth, and reproduction. Endothelial cells are ideally positioned to respond to hormones, which act in concert with locally produced chemical mediators to regulate their growth, motility, function, and survival. Hormones affect angiogenesis either directly through actions on endothelial cells or indirectly by regulating proangiogenic factors like vascular endothelial growth factor. Importantly, the local microenvironment of endothelial cells can determine the outcome of hormone action on angiogenesis. Members of the growth hormone/prolactin/placental lactogen, the renin-angiotensin, and the kallikrein-kinin systems that exert stimulatory effects on angiogenesis can acquire antiangiogenic properties after undergoing proteolytic cleavage. In view of the opposing effects of hormonal fragments and precursor molecules, the regulation of the proteases responsible for specific protein cleavage represents an efficient mechanism for balancing angiogenesis. This review presents an overview of the actions on angiogenesis of the above-mentioned peptide hormonal families and addresses how specific proteolysis alters the final outcome of these actions in the context of health and disease.
Blood vessels influence the metabolic effects of hormones by transporting fluid, nutrients, oxygen, and waste material. In addition, the vascular system delivers hormones from other parts of the body, allowing them to perform their local actions, which can include the production of another hormone that has to be transported to other specialized cells. Likewise, the function of blood vessels depends on hormones that regulate blood pressure, blood coagulation, and inflammation. To favor hormone delivery or blood transport into growing tissues, hormones promote angiogenesis, the proliferation of new blood vessels from preexisting vasculature. By acting systemically, hormones coordinate and integrate angiogenesis with other functions throughout the body. They also regulate blood vessel growth by controlling the production of local chemical mediators, often other hormones, but also growth factors, cytokines, enzymes, receptors, adhesion molecules, and metabolic factors, by vascular endothelial cells and other cells within the vicinity of capillaries. Notably, hormones can acquire angiogenic or antiangiogenic properties after undergoing proteolytic cleavage within the tissue microenvironment.
Proteolytic cleavage is a mechanism used frequently to generate proangiogenic and antiangiogenic protein mediators at specific sites. Proteolytic cleavage of extracellular matrix (ECM) components releases smaller proangiogenic fragments from larger proteins, as well as sequestered proangiogenic growth factors and cytokines (380). Similarly, antiangiogenic peptides are generated via proteolysis of components of the ECM and the coagulation and fibrinolytic systems (404, 519). Although proteolytically processed antiangiogenic fragments have long been known (160, 404), little attention has been given to proteolytic cleavage as an important mechanism controlling hormone action on angiogenesis.
Here, we review the regulation of angiogenesis by representative peptide hormones that are converted to proangiogenic or antiangiogenic molecules by proteolytic cleavage. The properties of these fragments versus those of their precursors, the regulation of the protease(s) responsible for specific protein cleavage, and the selective expression of specific receptors and associated signaling pathways for each hormonal isoform are discussed within a wider context of health and disease, with the expectation that understanding the role of hormones in angiogenesis could open new therapeutic perspectives for diseases resulting from angiogenic dysregulation.
II. ANGIOGENESIS OVERVIEW
The formation of new blood vessels is essential for organogenesis and successful embryonic and fetal development (139). In the adult organism, the proliferation of blood vessels is key for the growth and function of female reproductive organs, such as the ovary and endometrium during the menstrual cycle and the placenta and mammary gland during pregnancy (219). However, in most adult tissues, physiological angiogenesis is highly restricted, and capillary growth occurs only rarely and in association with repair processes such as wound and fracture healing. Disruption of the mechanisms controlling physiological angiogenesis has a major impact on health, as it underlies the pathogenesis of a growing list of diseases characterized by the overproliferation of blood vessels, including cancer, psoriasis, arthritis, retinopathies, obesity, asthma, and atherosclerosis. In addition, insufficient angiogenesis and abnormal vessel regression can lead to heart and brain ischemia, neurodegeneration, hypertension, osteoporosis, respiratory distress, preeclampsia, endometriosis, postpartum cardiomyopathy, and ovarian hyperstimulation syndrome (74, 75).
Hypoxia is an important stimulus for the physiological and pathological growth of blood vessels (449). It connects vascular oxygen supply to metabolic demand. Cells are normally oxygenated by diffused oxygen, but when tissues grow beyond the limit of oxygen diffusion, hypoxia triggers vessel growth by signaling through hypoxia-inducible transcription factors that upregulate and downregulate the expression of proangiogenic and antiangiogenic factors, respectively (449).
In the embryo, blood vessels originate from endothelial cell progenitors that migrate into avascular areas and form a primitive vessel network. This de novo formation of blood vessels, termed vasculogenesis, is followed by sprouting, branching, and stabilization in the process known as angiogenesis that utilizes existing vasculature to generate new vessels. Even though both vasculogenesis and angiogenesis have long been known to initiate a functional circulatory system during embryogenesis (267), that vasculogenesis also operates in the adult has been clarified only recently. In adults, endothelium-derived circulating cells and stem cell precursors of endothelial cells contribute to vessel growth in both physiological and pathological conditions (139, 451).
Vascular endothelial growth factor (VEGF) and angiopoietin-1 and -2 are probably the most important factors promoting angiogenesis, but the regulation of this process is complex and involves an extensive interplay between cells, multiple soluble factors, and ECM components (Fig. 1), which has been the focus of detailed reviews (5, 74, 267, 303). The following is a brief summary of the angiogenesis process illustrating some of the relevant molecular interactions.
Angiogenesis is stimulated by hypoxia, which upregulates the expression of several genes involved in different steps of angiogenesis, including VEGF, angiopoietin-2, and nitric oxide synthases (NOS) (449). Vessels dilate in response to nitric oxide (NO), and VEGF disrupts endothelial cell contacts causing vasopermeability (26). Endothelial cells are then free to migrate out of the basement membrane and through the softened perivascular space using leaked plasma proteins as a provisional matrix. However, only a subset of endothelial cells is selected for sprouting and respond to proangiogenic signals by mechanisms involving Notch receptors and their Delta-like 4 ligand (217). The directional migration of endothelial cells is primarily driven by VEGF, angiopoietin-1, angiopoietin-2, and basic fibroblast growth factor (bFGF). Other contributing cytokines include platelet-derived growth factor (PDGF), epidermal growth factor (EGF), transforming growth factor-β (TGF-β), and the guidance molecules ephrins, semaphorins, and netrins (5, 303). Proteases released by endothelial cells, including plasminogen activators, matrix metalloproteases (MMPs), heparinases, chymases, tryptases, and cathepsins, degrade and alter the composition of the ECM, allowing adequate support and guidance for endothelial cell migration. In addition to migration, sprout extension involves the proliferation of endothelial cells induced primarily by VEGF and angiopoietin-2 (in the presence of VEGF). Proteases facilitate vessel sprouting by releasing ECM-bound angiogenic activators (bFGF, VEGF, TGF-β) and activating angiogenic cytokines like interleukin (IL)-1β. The degraded ECM enables integrin-mediated endothelial cell adhesion and the signaling of integrins, such as αVβ3 and αVβ5, that participate in cross-talk with VEGF and bFGF receptors to either promote or inhibit endothelial cell proliferation (491). Eventually, the sprouts join together and convert into tubes first by intracellular and then intercellular fusion of large vacuoles (273). Vascular endothelial statin, a secreted ECM-associated protein, participates in lumen formation (427). The subsequently enhanced delivery of oxygen by the onset of blood flow lowers local VEGF expression, thereby reducing endothelial cell proliferation. These events together with the recruitment of pericytes and the deposition of a subendothelial basement membrane promote vessel maturation and quiescence. These processes are regulated by angiopoietin-1 and its receptor Tie 2, PDGF and PDGF receptor β, Notch signaling (25), sphingosine-1-phosphate-1 (S1P1)/endothelial differentiation sphingolipid G protein-coupled receptor (16), and TGF-β (53). Proteases can also modulate and finalize the angiogenesis process by liberating ECM-bound antiangiogenic factors such as thrombospondin-1, canstatin, arresten, tumstatin, and endostatin and non-matrix-derived inhibitors of vessel formation including angiostatin, vasoinhibins, antithrombin III, prothrombin kringle-2, plasminogen kringle-5, and vasostatin (404).
III. HORMONES WITH EFFECTS ON ANGIOGENESIS ALTERED BY PROTEOLYSIS
While tremendous efforts have been devoted to the study of factors that primarily regulate angiogenesis, the contributions of broadly acting agents like hormones remain more difficult to interpret due to the quantity and diversity of their targets and actions. Hormones represent an appropriate system for regulating the angiogenesis process throughout the body. Indeed, the endocrine nature of antiangiogenic hormones enables the efficient delivery required, for example, to maintain the quiescent state of blood vessels that normally exists in adult tissues. By acting directly on endothelial cells or indirectly by recruiting other cell types to produce angiogenesis regulators, proangiogenic hormones can help turn on the angiogenesis switch to promote the growth and metabolism of target tissues. Moreover, by interacting with local factors, hormones may regulate angiogenesis differentially at specific sites.
Multiple peptide hormones regulate angiogenesis by acting as either stimulatory or inhibitory factors (Table 1). Others contain, within the same molecule, the ability to exert distinct or even opposite effects on angiogenesis: proteolysis converts the original hormone to either proangiogenic or antiangiogenic peptides. The presence of opposing activities within the same hormonal precursor provides an efficient mechanism for fine-tuning angiogenesis and implies that specific proteolytic cleavage controls a shift toward a proangiogenic state or an angiostatic condition. Clearly, the selective expression of the receptors for each hormonal isoform and their related signaling pathways also play a critical role.
The following is an overview of the actions on angiogenesis of members of the growth hormone (GH)/prolactin (PRL)/placental lactogen (PL) family, the renin-angiotensin system (RAS), and the kallikrein/kinin system (KKS), all of which are regulated by proteolysis.
A. The Growth Hormone/Prolactin/Placental Lactogen Family
GH, PRL, and PL are produced in the anterior pituitary gland as well as in the uteroplacental unit and other nonpituitary sites. They are structurally and functionally related and evolved from a common ancestral gene some 400 million years ago (107, 364, 588). They vary between 22 and 23 kDa and comprise 190–200 amino acids organized in a four-α-helix configuration stabilized by two (GH and PL) or three (PRL) disulfide bonds (Fig. 2). The relationship between their structural similarities and biological properties is unclear. Human PL and GH are remarkably similar in amino acid sequence (86% homology), size (191 amino acids), and disulfide bond number and position, whereas human PRL has 199 amino acids and 3 disulfide bonds and shares only 25% sequence identity with the other two hormones (399). Nevertheless, all three human hormones are potent agonists of the PRL receptor (204, 399), but only GH activates the GH receptor (330, 591).
GH, PRL, and PL are proangiogenic and upon proteolytic cleavage are converted to peptides with potent inhibitory effects on blood vessel growth and function, forming a family recently named vasoinhibins (Fig. 2) (92).
1. Growth hormone
GH is best known as a major regulator of linear postnatal growth, energy metabolism, muscle expansion and performance, and immune function (88, 455, 577). GH acts by stimulating insulin-like growth factor I (IGF-I) production both systemically and locally, and also by directly activating the JAK/STAT pathway in target cells (304, 309). Accumulating evidence indicates that GH helps regulate vascular growth and function. GH receptors have been detected in blood vessels from different vascular beds (296, 326, 494, 550, 603) and in cultured endothelial cells (320, 555), where GH stimulates endothelial cell proliferation (478, 529) and tube formation (67, 178). The proangiogenic effects of GH have also been demonstrated in vivo. Treatment with GH increases the number of cerebral cortical arterioles in aging rats (515), augments VEGF expression and angiogenesis in the rat myocardium after infarction (302, 468), stimulates wound angiogenesis in diabetic rats (191) and in mice (553), and may enhance vascularization by promoting mobilization of bone marrow-derived endothelial progenitor cells into the bloodstream (134, 554). Consistent with these findings, the skin of adult, GH-deficient patients shows reduced capillary density and permeability that improves after treatment with GH (418), and GH-deficient adults and children have reduced retinal vascularization (238, 239).
The retina has been considered a major target for the proangiogenic actions of the GH/IGF-I axis (for a review, see Ref. 180). Proliferative diabetic retinopathy is rare in dwarfs who are deficient for GH and IGF-I (359). Lowering circulating GH with a somatostatin analog decreases ischemia-induced retinal neovascularization in mice (510), which is also reduced by blocking GH receptor expression with antisense oligonucleotides (610) or by genetic silencing of the GH receptor in transgenic dwarf mice (510). GH action on retinal angiogenesis is mediated primarily through circulating or locally produced IGF-I (87). Administration of IGF-I to somatostatin analog-treated mice normalizes serum IGF-I levels, but not GH levels, and promotes ischemia-induced retinal neovascularization (510). Furthermore, deletion of the IGF-I receptor in vascular endothelial cells partially protects against retinal neovascularization (297), and IGF-I antagonists interfere with the actions of VEGF (511), the essential cytokine mediating proliferative retinopathies (68). Indeed, IGF-I is necessary for maximal VEGF activation of the mitogen-activated protein kinase (MAPK) and Akt pathways in the retina (511), and IGF-I induces the expression of retinal VEGF by activating the phosphatidylinositol 3-kinase (PI3-K)/Akt, hypoxia inducible factor-1 (HIF-1), NFκB, and JNK/AP-1 pathways (443).
IGF-I may also mediate the proangiogenic actions of GH at other sites, as IGF-I receptors are widely expressed in endothelial cells, and IGF-I has been shown to stimulate angiogenesis in vivo and in vitro (for a review, see Ref. 129). However, some actions of GH on endothelial cells may be independent of IGF-I. For example, GH is unable to increase the transcription of IGF-I in endothelial cells. Instead, it promotes the expression and activity of endothelial NOS (eNOS) (555), and eNOS-derived NO stimulates vasorelaxation, vasopermeability, and angiogenesis (586). In addition, systemic (320) or local infusions of GH (390) acutely increase forearm blood flow and NO release in healthy humans without significantly raising plasma IGF-I levels or muscle IGF-I mRNA expression. While these observations argue in favor of an autonomous GH action mediated by NO, IGF-I is also vasoactive due to its activation of eNOS (for a review, see Ref. 129), and more information is required to clarify this issue. The relevance of the vascular actions of GH is emphasized by clinical data showing that patients with GH deficiency have increased risk of cardiovascular death (206). Loss of GH production in humans leads to increased peripheral resistance, reduced cardiac output, decreased blood flow in response to vasodilators, and reduced systemic NO levels, whereas GH replacement restores these responses to normal (58, 73, 391). Interestingly, some vasoconstrictive effects of GH via NO inhibition have also been reported in experimental settings (370).
The complexity of GH vascular effects is further illustrated by the fact that elevated GH levels are not always associated with angiogenesis. Overexpression of GH in transgenic mice (gigantic phenotype) does not result in increased retinal neovascularization (510), acromegaly has no correlation with retinopathy (36, 167), and long-term GH replacement therapy does not appear to increase the risk of retinopathy in children or adults (239). Interestingly, in the chick embryo chorioallantoic membrane, GH is proangiogenic only during the more differentiated, nongrowing state of blood vessels (211, 529), and not in their less differentiated, proliferative phase (529). Furthermore, treatment with GH does not stimulate the proliferation of some endothelial cells in culture (67, 478). These contrasting findings imply that GH has context-dependent vascular actions influenced by other angiogenic agents, including IGF-I, NO, and VEGF, and GH itself. GH is produced by endothelial cells, and endothelium-derived GH stimulates the proliferation, migration, survival, and capillary formation of endothelial cells in an autocrine manner (67). Therefore, the lack of action of exogenous GH may relate to endothelial cell GH receptors being already occupied by the endogenous hormone. Other important factors affecting GH actions on angiogenesis include its local proteolysis to vasoinhibins (see below) and the relative contribution of other related hormones. Human GH, the form used therapeutically and in most experimental work, has the ability to activate PRL receptors, which can also mediate proangiogenic signals.
2. Prolactin and placental lactogen
PRL was named for one of its first known functions, the initiation and maintenance of lactation, but this hormone is remarkably versatile, as it regulates various events in reproduction, osmoregulation, growth, energy metabolism, immune response, brain function, behavior (44, 59), and angiogenesis (96). In vivo studies show that treatment with PRL stimulates the proliferation of endothelial cells in the corpus luteum (195), in the testis (291), and in the myocardium (243). During pregnancy, PRL receptor deficiency interferes with the vascularization of the corpus luteum (220), and defective mammary gland development in PRL (249) and PRL receptor (419) null mice can be associated, in part, with subnormal neovascularization (96). However, several inconsistencies reflect the complexity of PRL action. Like GH, PRL is proangiogenic in the chick embryo chorioallantoic membrane assay only during the nongrowing stage of blood vessels (529), and not during their proliferative phase (94, 196, 529, 567). Also, PRL is unable to stimulate blood vessel growth in the corneal angiogenesis assay (567), and targeted disruption of the PRL gene is associated with highly vascularized pituitary tumors in aged mice (116). A major unresolved issue is whether PRL directly or indirectly stimulates endothelial cell proliferation. The PRL receptor has been detected, albeit at low levels, in some endothelial cells (360, 461, 567) but not in others (410), and the preponderance of evidence shows no mitogenic effect of PRL on cultured endothelial cells (94, 160, 410, 461, 529, 567). Nonetheless, lack of an effect could reflect the fact that endothelial cells produce and release PRL able to promote proliferation in an autocrine manner (93, 461). One study reported a direct effect of PRL on endothelial cell growth that was dependent on the expression of heme oxygenase-1 (346), an enzyme that promotes cell cycle progression and prevents apoptosis by producing bilirubin and carbon monoxide (65, 132).
On the other hand, PRL can stimulate angiogenesis indirectly by inducing the synthesis of proangiogenic factors in other cell types. PRL promotes VEGF or bFGF expression in decidual cells (517), immune cells (207, 345, 559), and mammary epithelial cells (207). The PRL-induced release of VEGF by mammary epithelial cells and the Nb2 lymphoma cell line (207) depends on the JAK2/MAPK/early growth response gene-1 pathway, whereas the activation of heme oxygenase-1 mediates PRL-stimulated VEGF transcription in macrophages (346). Evaluating PRL action is rendered more difficult because human GH and PL can also activate PRL receptors. In fact, PL signals through the PRL receptor and may be an important contributor to increased blood vessel growth during pregnancy. Like PRL, PL stimulates in vivo angiogenesis in the chick embryo chorioallantoic membrane during the nonproliferative stage of blood vessels and is unable to stimulate the in vitro proliferation of certain endothelial cells (529).
PRL can affect the function of blood vessels during injury and inflammation. PRL alters the cytoskeleton and adhesion properties of endothelial cell monolayers after mechanical injury (360) and promotes the infiltration of leukocytes (415, 547) via their integrin-mediated adhesion to vascular endothelial cells (371). Also, PRL can have stimulatory or inhibitory effects on vascular resistance, blood volume, and blood flow depending on the experimental model and condition (for a review, see Ref. 96). Importantly, the vascular actions of PRL and PL, like those of GH, can be counteracted by their proteolytic conversion to vasoinhibins.
Vasoinhibins are a family of peptides derived by proteolytic cleavage from PRL, GH, and PL that inhibit blood vessel dilation, permeability, growth, and survival (reviewed in Refs. 92, 96). Cleavage by various proteases occurs near or within the large loop connecting the third and fourth α-helices in all three hormones (Fig. 2). Cathepsin D (35, 438), MMPs (MMP-1, MMP-2, MMP-3, MMP-8, MMP-9, and MMP-13) (339), and bone morphogenic protein-1 metalloprotease (BMP-1) (196) cleave PRL to generate vasoinhibins, whereas thrombin and plasmin cleave PL (477, 529) and BMP-1, thrombin, plasmin, subtilisin, and chymotrypsin cleave GH (196, 319, 615). Since only one study has addressed the actions of vasoinhibins derived from GH and PL (529), most of the following information concerns vasoinhibins originating from PRL.
Vasoinhibins decrease angiogenesis in the chick embryo chorioallantoic membrane (94, 196, 529) and in the cornea (141); inhibit blood vessel growth and survival, vasodilation, and vasopermeability in the retina (22, 140, 192); impair growth and function of coronary vessels (243); and reduce the growth, metastasis, and neovascularization of tumors (49, 285, 398). Vasoinhibins act directly on endothelial cells to reduce the mitogenic effects of VEGF and bFGF (94, 529); inhibit bFGF- and IL-1β-stimulated endothelial cell migration (312, 313) and tube formation (94, 313); block VEGF-induced vasopermeability (192); and prevent VEGF-, bradykinin (BK)-, and acetylcholine-stimulated vasodilation (192, 209). Vasoinhibins also promote vessel regression by stimulating apoptosis in endothelial cells (351, 537). The receptors for vasoinhibins have not been identified (97), but portions of their signaling pathways have been defined (for a review, see Ref. 96). Vasoinhibins arrest endothelial cells at G0 and G2 by inhibiting cyclin D1 and cyclin B1 and stimulating the cyclin-dependent kinase inhibitors p21cip1 and p27kip1 (535). This regulation may involve the inhibition of MAPK activation (117, 118) and of eNOS-dependent proliferative pathways (209, 649). Vasoinhibins also interfere with endothelial cell migration by upregulating plasminogen activator inhibitor type-1 (312) and blocking the Ras-Tiam1-Rac1-Pak1 signaling pathway (313). Vasoinhibins inhibit endothelial cell survival by activating proapoptotic proteins of the Bcl-2 family (351) and the NFκB-mediated activation of caspases (537). Finally, vasoinhibins block vasodilation and vasopermeability by interfering with the Ca2+-dependent activation of eNOS (209) and by activating protein phosphatase 2A, which dephosphorylates and inactivates eNOS (192).
The extensive influence of vasoinhibins on vascular endothelial cells includes proinflammatory actions. Vasoinhibins act on endothelial cells to promote the expression of cell adhesion molecules (ICAM-1, VCAM-1, E-selectin) and of chemokines from the CXC and the CC families (536). Consequently, they stimulate the NFκB-mediated adhesion of leukocytes to endothelial cells and the infiltration of leukocytes into tumors (536). Also, vasoinhibins act as proinflammatory mediators on other cell types, like pulmonary fibroblasts, where they promote the NFκB-mediated expression of inducible NOS (iNOS) with potency comparable to the combination of IL-1β, tumor necrosis factor (TNF)-α, and interferon (IFN)-γ (110, 340).
Vasoinhibins are emerging as natural inhibitors of the angiogenic process. They have been identified in the retina (22) and cartilage (339), where angiogenesis is highly restricted, and blocking the expression and action of vasoinhibins by siRNA targeting PRL or neutralization with antibodies results in stimulation of retinal angiogenesis and vasodilation (22). In addition, interfering with the formation of vasoinhibins by pharmacologically blocking pituitary PRL secretion prevents postpartum cardiomyopathy in mice (243). In chondrocytes, vasoinhibins appear to be generated mainly by MMPs (339), whereas other proteases predominate at different sites. Genetic deletion of two genes encoding BMP-1-like metalloproteases prevents the generation of GH- and PRL-derived vasoinhibins in mouse fibroblasts (196). Recently, cathepsin D has been shown to generate vasoinhibins within PRL secretory granules in the anterior pituitary, suggesting that vasoinhibins are released during the process of exocytosis (M. Cruz-Soto and C. Clapp, unpublished observations). Notably, vasoinhibins were not detected in the anterior pituitary glands of cathepsin D-null mice, arguing in favor of cathepsin D being the physiologically relevant protease at this site (Cruz-Soto and Clapp, unpublished observations). Understanding the mechanisms regulating the generation of pituitary vasoinhibins could help clarify conflicting data, such as the presence of highly vascularized pituitary tumors in PRL-null mice (116). Indeed, because the pituitary gland may be an important site for the cleavage of PRL into vasoinhibins (92), it is possible that the absence of vasoinhibins, rather than the lack of PRL, accounts for the upregulation of angiogenesis in pituitary tumors.
Vasoinhibins are not the only members of the PRL/GH/PL family that inhibit angiogenesis. S179D PRL, a molecular mimic of naturally occurring phosphorylated PRL, was recently shown to inhibit angiogenesis by interfering with endothelial cell migration, proliferation, survival, and growth factor signaling (567, 568). Phosphorylation of PRL at Ser-179 alters the charge of the molecule, and it will be useful to examine the structural properties of phosphorylated PRL for common features with vasoinhibins.
B. The Renin-Angiotensin System
The RAS encompasses various peptide hormones that act systemically and locally to control blood pressure and body fluid homeostasis. Over the last two decades, RAS has been a key target for the development of drugs effective for the treatment of cardiovascular diseases, including hypertension, renal diseases, cardiac hypertrophy, congestive heart failure, and ischemic heart disease (235, 349, 602).
RAS members result from stepwise enzymatic processing (Fig. 3) that starts with the cleavage of circulating angiotensinogen (AGT) by the aspartyl protease renin, forming the inactive decapeptide angiotensin I (ANG I) and the COOH-terminal protein des[ANG I]AGT. Subsequently, angiotensin-converting enzyme (ACE) cleaves ANG I to generate the main effector of RAS, the octapeptide angiotensin II (ANG II). Other endopeptidases, like neprilysin, prolylcarboxypeptidase (PrCP), and angiotensin converting enzyme-related carboxypeptidase (ACE2), cleave ANG I and ANG II to generate the heptapeptide ANG-(1–7) (311), while ACE2 also converts ANG I into the nonapeptide ANG-(1–9). Systemic ANG II is primarily produced within the pulmonary circulation, and its formation is limited by the availability of circulating renin released by the juxtaglomerular cells of the kidney. All components of RAS are present in many tissues, where ANG I and ANG II can also be produced by enzymes other than renin and ACE, including tonin, cathepsin G, chymostatin-sensitive ANG II-generating enzyme, chymase, and tissue plasminogen activator (Fig. 3) (311). Of importance, ACE can cleave ANG-(1–9) to ANG-(1–7), and ANG-(1–7) to ANG-(1–5), while neprilysin, chymase, and aminopeptidase A inactivate ANG II.
Increasing evidence indicates that members of RAS exert both positive and negative effects on angiogenesis. Because the main role of RAS is to maintain body fluid homeostasis in response to a drop in perfusion pressure, RAS hormones likely participate in controlling neovascularization during vasoconstriction-associated ischemia. Intricate mechanisms govern their contrasting actions on angiogenesis and involve the activation of different receptor subtypes with opposite outcomes depending on specific tissue and disease conditions as well as the proteolytic conversion of larger precursor molecules, which may or may not have angiogenic actions, into smaller angiogenic or antiangiogenic peptides.
1. Angiotensin II
The most fully characterized component of the RAS system, ANG II, is normally proangiogenic when administered in vivo or in vitro (for reviews, see Refs. 236, 276). ANG II stimulates vessel proliferation in the chick embryo chorioallantoic membrane (308), the rabbit cornea (158), and the Matrigel model in mice (544). The two G protein-coupled receptors for ANG II, AT1 and AT2, have been identified in endothelial cells (126), and AT1 is the main mediator of the proangiogenic actions of ANG II. Inhibitors of AT1 block the proangiogenic effect of ANG II in the mouse Matrigel model (544), the rat ischemic hindlimb (545), and the electrically stimulated rat skeletal muscle (18). Likewise, inhibitors of AT1, but not of AT2, prevent the stimulatory effects of ANG II on the proliferation and tube formation of cultured endothelial cells (421, 526) and bone marrow-derived endothelial progenitor cells (257). Also, AT1 antagonists inhibit the ANG II-induced upregulation of VEGF, VEGFR2, angiopoietin-2, and Tie2 in endothelial cells (90, 421) and in vascular smooth muscle cells (462). However, in kidney, both AT1 and AT2 inhibitors are effective (465). ANG II upregulates other proangiogenic mediators, including NO (416, 544), bFGF (431, 518), PDGF (106, 386), IGF-I and its receptor (63, 227, 581), EGF (186), hepatocyte growth factor (HGF) (153), and TGF-β (198, 271, 286, 414). ANG II can also directly signal endothelial cell proliferation via c-Src-mediated increase of NADP oxidase activation (563), phosphorylation of ERK1/2 (429), p38 MAP kinase (571), and JNK (488), as well as reduced phosphorylation of Src homology 2-containing inositol phosphatase 2 (SHP-2) (482).
On the other hand, ANG II is antiangiogenic under certain conditions, as inhibition of endogenous ANG II production or action by treatment with ACE inhibitors or AT1 and AT2 blockers, respectively, stimulates angiogenesis in vivo (236, 593). ANG II may inhibit angiogenesis by activating the AT2 receptor. AT2 expression is augmented in the ischemic limb of wild-type mice, and vessel density increases in the ischemic limb of AT2-knockout mice (505). Furthermore, ANG II-induced neovascularization in the cremasteric muscle is enhanced by AT2 blockers, although it is inhibited by AT1 antagonists (384). However, the contribution of each receptor subtype to the antiangiogenic effect of ANG II appears to depend on the chosen angiogenesis model. In the alginate tumor model, AT2-null mice show impaired angiogenesis, and treatment with AT1 inhibitors promotes angiogenesis (593). Interestingly, evidence for the antiangiogenic effects of ANG II via AT2 has also been provided in cultured endothelial cells (47), where AT2 receptor stimulation inhibits VEGF-induced endothelial cell migration and tube formation.
Although much needs to be learned about the mechanisms controlling the dual effects on angiogenesis of AT1 and AT2, it is generally accepted that the two receptors essentially mediate contrasting effects of ANG II (for reviews, see Refs. 124, 276). AT1 receptors are ubiquitously expressed and responsible for most of the well-known actions of ANG II, including vasoconstriction, aldosterone and vasopressin release, renal sodium and water reabsorption, sympathetic activation, augmented cardiac contractility, smooth muscle cell proliferation, vascular and cardiac hypertrophy, inflammation, and oxidative stress. In contrast, AT2 receptors are restricted to organs like the brain, kidney, adrenals, uterus, ovary, and the cardiovascular system, where their activation leads to vasodilation, lower blood pressure, reduced cardiac and vascular hypertrophy, anti-inflammation, and suppressed growth, tissue repair, and apoptosis. Each receptor subtype activates multiple signaling pathways (reviewed in Refs. 153, 276), and the opposing influence of the two receptor subtypes is supported by signaling experiments in vitro. For example, AT1 receptors upregulate the expression of VEGF and angiopoietin-2 in microvascular endothelial cells by stimulating the release of heparin-binding EGF followed by the transactivation of the EGF receptor, whereas AT2 attenuates these actions by blocking EGF receptor phosphorylation (186). Also, AT1 acts in endothelial cells through the PI3-K/Akt pathway to upregulate survivin, suppress caspase-3 activity (412), and induce focal adhesion kinase/paxillin phosphorylation (372), which lead to endothelial cell survival and migration, respectively. In contrast, AT2 blocks VEGF-induced phosphorylation of Akt, causing the inhibition of eNOS activation, endothelial cell migration, and tube formation (47). ANG II regulation of endothelium-derived NO is complicated, as both positive and negative effects of AT1 and AT2 have been reported (42, 416, 645). AT2 signals vasodilation directly by promoting eNOS activity and endothelial NO release (416), but also indirectly by stimulating the production of BK, which in turn stimulates the eNOS/NO/cGMP pathway in the endothelium (564). These AT2 actions counterbalance the vasoconstriction elicited by AT1-induced inhibition of eNOS (416), G protein-mediated activation of phospholipase C (PLC), and inositol trisphosphate-induced mobilization of intracellular Ca2+ in vascular smooth muscle cells (124). Other contrasting mechanisms are the AT1 activation of tyrosine kinases that phosphorylate and activate the Ras/Raf/MAPK cascade, resulting in cellular growth and survival, and the AT2 activation of MAPK phosphatase-1 that inactivates ERK1 and ERK2, leading to Bcl-2 dephosphorylation and Bax upregulation (246).
Further complexity in the regulation of angiogenesis by RAS hormones is illustrated by the fact that AGT, des[ANG I]AGT, and ANG-(1–7) are antiangiogenic.
2. Angiotensinogen and des[ANG I]angiotensinogen
The human AGT protein contains 452 amino acids; the first 10 correspond to ANG I and the remainder to des[ANG I]AGT (Fig. 3). Circulating AGT is synthesized in the liver, but AGT is also produced in tissues such as the brain, large arteries, kidney, heart, and adipose tissue, where it is hydrolyzed by extravascular renin or other proteases to ANG I or directly to ANG II (Fig. 3) (111). AGT shares structural homology with the serine protease inhibitor (serpin) family of proteins, and the fact that some serpin proteins regulate angiogenesis [angiogenin (190), pigment epithelium-derived factor (PEDF) (123), maspin (30), and cleaved antithrombin (406)] led to the discovery of the antiangiogenic properties of AGT and des[ANG I]AGT. Both proteins inhibit in vivo angiogenesis in the chick embryo chorioallantoic membrane and the in vitro proliferation, migration, and tube formation of endothelial cells (20, 80). No specific AGT or des[ANG I]AGT receptors have been detected, but both proteins bind to AT1 and AT2 receptors at their micromolar plasma concentrations (197). AGT does not affect the expression of angiogenic growth factors in vivo; instead, it directly suppresses proliferation of endothelial cells and induces their apoptosis (62). AGT-deficient mice exposed to cold display an abnormal vascular brain barrier phenotype (272) that may reflect blood vessels destabilized by the absence of AGT in glial cells surrounding brain capillaries (527).
Local conditions may determine whether the antiangiogenic properties of AGT and des[ANG I]AGT prevail over the proangiogenic effects of ANG II. For example, circulating AGT could prevent angiogenesis at sites lacking renin.
ANG-(1–7) is present in the circulation and in tissues, where its concentration increases after any condition that raises plasma or tissue levels of ANG I (161). For example, circulating levels of ANG-(1–7) increase 25- to 50-fold during ACE inhibition (71, 294, 377) due to increased ANG I conversion to ANG-(1–7) and inhibition of ANG-(1–7) breakdown by ACE (83). Although ANG I is the primary substrate for the generation of ANG-(1–7), the latter may also be formed from ANG II (Fig. 3), and higher availability of ANG II after treatment with blockers of the AT1 receptor also elevates circulating ANG-(1–7) (71, 294, 377).
ANG-(1–7) was characterized as the first NH2-terminal angiotensin peptide opposing the vasopressor, proliferative, and angiogenic actions of ANG II, thereby endowing RAS with greater capability for regulating tissue perfusion (161, 341, 528). ANG-(1–7) acts as a vasodilator in vascular beds of different species (161, 162, 299, 458, 486), lowers blood pressure (48), reduces the proliferation of smooth muscle vascular cells in vitro (176) and in vivo (528), is cardioprotective (483), and inhibits angiogenesis and fibrovascular tissue infiltration in the mouse sponge model (341).
Although there is some evidence that ANG-(1–7) can activate ANG II receptors (592), the G protein-coupled Mas receptor has been identified as the functional receptor for ANG-(1–7) (12). Mas is expressed in endothelial cells (119), and its targeted disruption in mice causes increased blood pressure, endothelial dysfunction, imbalance between NO and reactive oxygen species, and a major cardiovascular phenotype, all of which are consistent with lack of ANG-(1–7) signaling (625). ANG-(1–7) induces endothelium-dependent vasodilation by stimulating NO release indirectly through BK-induced eNOS activation (237, 299), but also directly by promoting Mas-mediated eNOS activation via Akt (482). In addition, ANG-(1–7) inhibits ANG II-induced phosphorylation of MAPK through prostacyclin-mediated production of cAMP and activation of cAMP-dependent kinase (542), and it prevents ANG II-induced activation of Src and its downstream targets ERK1/2 and NADPH oxidase (482). ANG-(1–7) induces as well the phosphorylation of SHP-2, which could act as a negative regulator of ANG II-induced MAPKK and Src signaling (482).
In summary, due to the dual effects of RAS hormones, local conditions affecting their synthesis and clearance rate, the production and activity of the converting proteases, and their receptor-mediated signaling pathways would determine whether an antiangiogenic or an angiogenic condition prevails. ACE inhibitors and AT1 receptor blockers, which are among the most widely prescribed drugs for blood pressure control, lower ANG II generation and action and increase renin levels, accelerating AGT cleavage into antiangiogenic des[ANG I]AGT and ANG-(1–7) (311). However, treatment with ACE inhibitors is frequently proangiogenic (for a review, see Refs. 236, 504), primarily due to blockage of ACE-induced degradation of the potent vasodilator and proangiogenic hormone BK (236, 504). RAS/BK interactions are essential for balancing blood pressure and illustrate the sophisticated local and systemic interactions influencing the final angiogenic response.
C. The Plasma Kallikrein-Kinin System
The KKS in plasma comprises the serine proteases, coagulation factor XII and plasma prekallikrein, and high-molecular-weight kininogen (HK). The plasma KKS is known as the contact system, because originally the only known mechanism for its activation was contact with artificial, negatively charged surfaces. However, it is now recognized that the contact system is activated physiologically at the cell membrane, leading to the generation of small peptides and proteins with effects on blood coagulation, inflammation, pain, natriuresis, blood pressure, vascular permeability, and angiogenesis (for reviews, see Refs. 8, 481, 487).
HK is a 120-kDa, single-chain glycoprotein composed of six domains (D1 to D6), originally identified as the protein that, upon cleavage, yields the nonapeptide BK, the main effector of the plasma KKS (266) (Fig. 4). Most plasma prekallikrein circulates bound to HK, and on the endothelial cell surface HK serves as its main binding site (381). Upon binding to the endothelial cell surface via HK, prekallikrein is converted to kallikrein by prolylcarboxypeptidase (PrCP) (495), the same serine protease that generates ANG-(1–7) and ANG II (Fig. 3). Kallikrein favors factor XII association with and activation by endothelial cells, and activated factor XII (factor XIIa) feeds back to cleave prekallikrein to kallikrein, accelerating its formation (for a review, see Ref. 487). Kallikrein cleaves D4 in HK to release BK and cleaved HK (HKa), a two-chain structure (D1-D3 and D5-D6) linked through a single disulfide bond (Fig. 4). HK is also converted to BK and HKa by factor XIIa, albeit to a lesser extent than by plasma kallikrein (102).
Various peptidases, including ACE, metabolize BK to generate smaller peptides that have no known effect on angiogenesis (70). Pharmacological inhibitors of ACE, both by reducing ANG II levels and by blocking the degradation of BK, help control cardiovascular diseases, including hypertension and myocardial infarction (for reviews, see Refs. 236, 604). In fact, hypertension and cardiac failure have been functionally linked to impaired angiogenesis (236, 284, 604), and the plasma KKS is another hormonal system in which a protein generates both proangiogenic (BK) and antiangiogenic (HKa) fragments after undergoing proteolytic cleavage (Fig. 4).
Discovered half a century ago, BK has multiple biological activities (reviewed in Ref. 348) and is involved in pathological states including hypertension (278) and inflammation (481). Besides being a well-known vasodilator and vasopermeability factor, BK has clear proangiogenic effects (236). It stimulates the proliferation, tube formation, and survival of cultured endothelial cells (366, 376) and promotes angiogenesis in the rabbit cornea (426), the chick embryo chorioallantoic membrane (104), the nude mouse xenograft assay (514), the rat subcutaneous-sponge model (234, 251), and the mouse ischemic hindlimb (146, 147). Moreover, angiogenesis is suppressed in HK-deficient rats and is restored by treatment with a BK analog (234).
BK stimulates angiogenesis by activating two G protein-coupled receptor subtypes, B1 and B2 (see Refs. 8, 356). Both receptors are present in endothelial cells (614), but while the B2 receptor is constitutively expressed, the B1 receptor is upregulated following tissue damage, ischemia, and inflammation (see review in Ref. 8). Actually, activation of B1 receptors may be seen as a mechanism to magnify BK actions, since this receptor subtype is normally expressed in the same cell types as the B2 receptor, uses similar signaling pathways but is less vulnerable to desensitization, and mediates the activity of other kinin metabolites (8). Inhibitors of receptors B1 (234, 321) and B2 (234, 262, 489) reduce in vivo angiogenesis and inhibit BK-induced endothelial cell proliferation (366, 376) and tube formation (366) in vitro. Notably, BK also stimulates endothelial cell growth and permeability by increasing B2 receptor-mediated VEGF expression in fibroblasts (255, 262) and smooth muscle cells (289), whereas the B1 receptor contributes to the proangiogenic action of BK by increasing bFGF synthesis in endothelial cells (137, 376, 426).
The mechanisms mediating the vascular effects of BK are not entirely understood, but it is clear that eNOS-derived NO is a major effector of BK-induced vasodilation (51, 209), vasopermeability (439), and angiogenesis (321, 426, 556). Although both B1 (426) and B2 receptors (366) signal to promote eNOS activity, B2 receptors likely predominate in mediating this action. BK stimulation of B2 receptors on endothelial cells activates PLC and phospholipase A2, which in turn trigger the intracellular mobilization of Ca2+ and the activation of eNOS via calmodulin binding. In addition, BK stimulates the activation of eNOS by promoting PI3-K/Akt-induced eNOS phosphorylation at Ser-617, Ser-635, and Ser-1179 and by stimulating calcineurin-mediated eNOS dephosphorylation at Thr-497, modifications that serve to increase the Ca2+-calmodulin sensitivity of the enzyme (for a review, see Ref. 579). BK also induces the phosphorylation/activation of VEGFR2, which can promote eNOS activation (366, 556).
Of importance, BK is rapidly inactivated in the intravascular compartment (half-life ∼15 to 30 s; Ref. 378), and it disappears completely after a single passage through the pulmonary circulation. Its degradation is ensured by three different kininases: ACE, aminopeptidase P, and carboxypeptidase N, which cleave at positions 7–8, 1–2, and 8–9 of BK, respectively. The stable metabolic end product of BK produced by ACE is a pentapeptide, known as BK-(1–5) (Fig. 4), that inhibits thrombin-induced platelet aggregation (385) but has no known effect on angiogenesis. BK inactivation and ANG I to ANG II conversion in the lungs are caused by the same ACE enzyme (397), and because ACE has a higher affinity for BK than for ANG I (269), the interplay between RAS and KKS must be considered to fully understand the role of both systems in angiogenesis.
Blockage of ACE stimulates angiogenesis in the heart of stroke-prone, spontaneously hypertensive rats (205) and in the rat limb muscle (69), promotes ischemia-induced angiogenesis in the rabbit hindlimb (155), and induces pseudocapillary formation and endothelial cell growth in vitro (137). The effect of BK after ACE inhibition is illustrated by the fact that neovascularization is greater in ischemic wild-type mice treated with ACE inhibitors than in the corresponding B2-receptor-null animals (504). The proangiogenic effect of ACE inhibition in the ischemic hindlimb is suppressed in B2-receptor-deficient mice (143). Importantly, the RAS and KKS interaction can also regulate angiogenesis via ANG II-dependent activation of B2 receptors (1, 383, 512, 564, 607), which leads to proangiogenic and vasodilator effects. The impact of ACE activity on RAS and BK generation and how it relates to angiogenesis regulation has been recently reviewed (236).
Predicting the outcome of KKS hormone actions on angiogenesis is complicated by the concomitant production of proangiogenic BK and antiangiogenic HKa.
2. Cleaved high-molecular-weight kininogen
HKa encompasses one heavy chain composed of D1 to D3 and a light chain corresponding to D5 and D6 linked by a disulfide bond (Fig. 4). The conversion of HK to HKa involves important conformational changes that expose a major region of D5 (599), which mediates the acquired ability of HKa to inhibit angiogenesis. HKa and recombinant D5 (also named kininostatin, Ref. 103) inhibit proliferation, migration, and survival of cultured endothelial cells (103, 226, 277, 640). Inhibition of endothelial cell proliferation is Zn2+ dependent and interferes with the mitogenic effects of bFGF, VEGF, HGF, and PDGF (640) by blocking the transition from G1 to S phase of the cell cycle (226). HKa also inhibits bFGF-induced angiogenesis in the Matrigel plug assay in mice and the cornea of rats (640), and both HKa and recombinant D5 inhibit the proangiogenic effects of bFGF and VEGF in the chick embryo chorioallantoic membrane (103, 640). The antiangiogenic properties of HKa may be mediated by its interaction with urokinase receptors through the D5 region, an interaction that would compete with the urokinase plasminogen activator receptor (uPAR) for the binding of ECM proteins (vitronectin) (72). An alternative mechanism is the binding of HKa to the cytoskeletal protein tropomyosin (641), which is involved in mediating the antiangiogenic properties of endostatin, an antiangiogenic fragment of collagen XVIII (338). The proposed signaling pathways underlying the antiangiogenic effects of HKa through its D5 region include the disruption of focal adhesions via vitronectin-uPAR-αvβ3-caveolin-Src-FAK-paxillin (102) or by vitronectin-uPAR-PI3K/Akt-paxillin (277), the reduction of cyclin D1 expression (226), the activation of Cdc2 kinase/cyclin A (597), and the generation of reactive oxygen species dependent on ECM components (531).
In summary, specific proteolytic cleavage of HK yields both proangiogenic (BK) and antiangiogenic (HKa) moieties. Because BK is hydrolyzed very rapidly and thus is active only close to its site of formation, BK may function as an initiator of angiogenesis, particularly in inflammatory and ischemic conditions (356), when the induction of B1 receptors can amplify its action. In contrast, HKa has a long in vivo half-life (9 h) and may counteract the angiogenic response to BK with time, tipping the balance towards antiangiogenesis. Of note, the outcome of the plasma KKS effect on angiogenesis depends not only on a repertoire of endothelial cell membrane proteins (i.e., HK-binding proteins, bound proteases, and receptors) but also on ECM proteins. HKa only induces death of proliferating endothelial cells on permissive surfaces like gelatin, fibronectin, vitronectin, and laminin (640), and its effect on vessel regression depends on the type of collagen (531). Furthermore, compensatory actions of RAS peptides may be integrated in the regulation of angiogenesis by the KKS, since the same enzyme (ACE) upregulates and downregulates the accumulation of proangiogenic ANG II and BK, respectively.
IV. HORMONAL REGULATION OF PHYSIOLOGICAL ANGIOGENESIS
Angiogenesis is normally absent in most adult organs except in female reproductive organs where intense vascular growth and regression occur physiologically, contributing to their growth, function, and involution. Members of the GH/PRL/PL family, the RAS, and the KKS regulate the physiology of reproductive organs, and some of their actions involve the control of blood vessel growth and regression.
A. Female Reproductive Organs
The endocrine system orchestrates a series of cyclical events in the female reproductive organs, allowing the ovum to mature and eventually become fertilized and the resulting new individual to be nurtured during the intrauterine and neonatal periods. Gonadotropin hormones from the pituitary gland drive ovarian function by stimulating estrogen and progesterone production, follicle growth, ovulation, and corpus luteum development. After fertilization, the maintenance of the corpus luteum is dependent on chorionic gonadotropin and PRL, permitting the continued secretion of progesterone necessary to maintain the endometrium in a state favorable for implantation and placentation (for reviews, see Refs. 27, 525). As for the growth, differentiation, and secretory activity of the mammary gland during pregnancy and lactation, the relative contribution of PRL, GH, and PL varies among species (44), but activation of the PRL receptor is crucial (396).
In addition to these hormonal inputs, evidence indicates that GH, directly or via stimulation of IGF-I production, regulates follicular growth, sexual maturation, and luteal function (28, 636) and that ANG II produced in response to gonadotropins can stimulate follicular maturation, steroidogenesis, ovulation, and corpus luteum growth and regression (see review, Ref. 632). Also, BK has been implicated in ovulation (546, 633), and both the RAS (400) and the KKS (99) participate in regulating implantation and placentation.
Reproductive events are also determined by cyclical changes in blood vessel growth and regression within the various reproductive organs, driven primarily by VEGF and then by its blockage (for reviews, see Refs. 173, 174, 202). The transient interruption of VEGF signaling by agents specifically designed to inhibit VEGF or block its receptors suppresses follicular development, ovulation (620, 623), corpus luteum formation, progesterone release (622), and postmenstrual endometrial regeneration (156). Homozygous and heterozygous knockouts of the VEGF gene exhibit major defects in placental blood vessels (76, 159), and interference with placental VEGF compromises normal angiogenesis and leads to a poorly perfused fetoplacental unit (355). In addition, inactivation of VEGF severely limits the development and function of the mammary gland (472). The fact that female reproductive organs are under the control of hormones able to affect angiogenesis, either directly or via the expression and action of VEGF or other proangiogenic signals, strongly argues in favor of angiogenesis regulation as an essential hormonal function. We refer to recent reviews on the action of gonadotropins and steroid hormones that are considered to be important regulators of angiogenesis in reproductive organs (11, 440, 457) and keep our focus on certain proteolysis-derived peptide hormones.
The information regarding ovarian angiogenesis has been extensively reviewed (172, 173, 175, 543), and accumulating evidence shows that ovarian VEGF is under hormonal control. VEGF increases in granulosa and theca cells of follicles as they become dependent on gonadotropin stimulation (501), and inhibition of gonadotropin release using a gonadotropin releasing hormone (GnRH) antagonist inhibits VEGF expression and angiogenesis in ovulatory follicles (549). Other peptide hormones stimulate follicle VEGF expression and angiogenesis. For example, IGF-I and IGF-II promote VEGF expression in granulosa cells (350), and the PRL present in follicular fluid stimulates endothelial cell proliferation (78). Of interest, a functional angiotensin system exists in the endothelial cells of the early corpus luteum (292), and ANG II induces VEGF synthesis in endothelial cells (90), promotes the expression of bFGF in bovine luteal cells (524), and may contribute to LH-induced corpus luteum angiogenesis. Furthermore, LH upregulates ovarian ANG II production (see review, Ref. 632), and ANG II antagonists can block LH-induced expression of bFGF in luteal cells (524). In addition, bFGF and VEGF upregulate luteal ANG II secretion, which supports the mechanism promoting progesterone secretion in the early corpus luteum (292).
As the corpus luteum ages in a nonfertile cycle, VEGF levels decline (422), and the angiopoietin-2/angiopoietin-1 ratio rises (203), leading to the endothelial cell apoptosis and vascular breakdown characteristic of luteolysis. In contrast, in a fertile cycle, the corpus luteum is rescued by chorionic gonadotropin, which promotes endothelial cell survival by upregulating VEGF, angiopoietin-2, and Tie-2 (621). Another important hormone regulating luteal function is PRL. In rodents, the proestrus surge of circulating PRL in a nonfertile cycle induces luteolysis, but after pseudopregnancy or pregnancy, the increase in systemic PRL promotes the survival and function of the corpus luteum (194). The opposing actions of PRL are perplexing and may involve several mechanisms (525), including both proangiogenic and antiangiogenic effects. Treatment with PRL induces endothelial cell proliferation in the corpus luteum of cycling rats, whereas reducing circulating PRL levels with bromocriptine (195) or deleting the PRL receptor gene (220) interferes with corpus luteum angiogenesis. In addition, the levels of PRL-derived vasoinhibins may rise at the end of a nonfertile cycle, because PRL increases MMP-2 activity during PRL-induced luteal regression (150), and MMP-2 generates vasoinhibins from PRL (339). Moreover, at the onset of luteal regression, corpus luteum-derived endothelial and granulosa cells produce and release higher levels of the vasoinhibin-generating protease cathepsin D (152).
2. Uterine endometrium and placenta
Angiogenesis is required for the cyclic processes of endometrial growth, breakdown, and repair during the menstrual cycle, and it provides a richly vascularized tissue receptive for implantation and placentation (reviewed in Refs. 84, 202, 265, 560, 619).
Uterine angiogenesis is regulated by multiple hormones, among which estrogen and progesterone predominate in regulating endometrial growth and differentiation. There is good evidence that estrogen drives VEGF production and angiogenesis during the proliferative phase of the cycle, whereas the presence and the absence of progesterone have been implicated in endometrial angiogenesis during the secretory and postmenstrual phases of the cycle, respectively (see reviews in Refs. 11, 202, 265). Other hormones involved in endometrial angiogenesis include chorionic gonadotropin (457), adrenomedullin (401), relaxin (208), and two subjects of this review, PRL (264) and ANG II (646).
Systemic PRL derived from the anterior pituitary gland increases during proestrus and early pregnancy in rodents, whereas in humans, circulating PRL remains low during the menstrual cycle and gradually increases during gestation, reaching its maximum level at term (reviewed in Ref. 44). PRL is also produced by the decidua (264, 447), a specialized endometrial stromal tissue that differentiates in the luteal phase of the menstrual cycle and throughout pregnancy. The location and temporal presence of PRL suggest its influence on endometrial angiogenesis during the secretory phase of the cycle and at the time of implantation and early placental development (264). Consistent with this, pharmacologically induced hyperprolactinemia enhances endometrial thickness (471) and uterine gland hyperplasia (281). PRL receptors are localized in the decidua, cytotrophoblasts, synciotriotrophoblasts (337), differentiated stromal, and uterine natural killer cells (224), which are important cellular sources of proangiogenic factors like VEGF, placental growth factor, bFGF, and angiopoietins (84, 130, 322), and PRL stimulates the expression of bFGF by decidual cells (517). Also, the placenta produces GH and PL, both of which promote endometrial gland proliferation (403, 516) and can activate PRL receptors.
On the other hand, cyclic changes in the activity of the RAS also occur during the reproductive cycle and are reflected in the circulation and in the expression of all of its components in the uteroplacental unit (reviewed in Ref. 400). During the luteal phase, ANG II increases in stromal cells near endometrial spiral arterioles, and ANG II receptor expression is increased in endometrial glands and blood vessels (9). Because ANG II stimulates the contraction of uterine blood vessels (444), ANG II may contribute to the vasopressor mechanism that initiates menstruation, causing hypoxia-induced regression and degradation of upper endometrial tissue in response to the withdrawal of progesterone (115). Also, a role for ANG II in endometrium regeneration is suggested by the presence of ANG II and ANG II receptors in endometrial glandular and stromal cells during the proliferative phase (9).
During pregnancy, all the components of RAS are active in the placenta, and increased ANG II levels in the maternal circulation are associated with placental angiogenesis and elevated blood flow (for a review, see Ref. 646). AT1 receptors are expressed predominantly in placental blood vessels and mediate ANG II-induced proliferation (55) and NO production by endothelial cells from fetoplacental arteries (646). In addition to mediating proangiogenic actions, ANG II-induced NO production helps attenuate the vasopressive response to ANG II (646). This effect is relevant, as vasodilation of maternal systemic circulation and the increased blood flow within the fetoplacental unit are major adaptations of mammalian pregnancy (509). The vasodilator response to ANG-(1–7) is enhanced during gestation (395), and systemic and renal ANG-(1–7) levels increase during late pregnancy (64, 361, 573). The different components of the KKS have also been identified in the cycling (98) and early pregnant uterus (15), suggesting that BK and its receptors actively participate in increased uterine blood flow, vasopermeability, and angiogenesis.
In addition to proangiogenic mechanisms, antiangiogenic events operate to regulate the regional distribution of placental neovascularization (reviewed in Ref. 254). Antiangiogenic mechanisms maintain the avascular nature of the decidual layer immediately surrounding the synciotrophoblastic mass of the invading embryo. Antiangiogenic events, such as endothelial cell apoptosis, also occur during human spiral arterial remodeling, which prevents maternal vessels from invading the embryonic compartment and fetal vessels from growing into the uterus. Notably, placental trophoblasts produce the secreted form of the VEGF receptor 1 (sVEGFR1), also known as soluble fms-like tyrosine kinase-1 (sFlt-1), which functions as an endogenous trap for VEGF (38, 280) and is a major contributor to decidual avascularity (254). Proteolytically modified hormones may participate in inhibiting placental angiogenesis as well. Cathepsin D is produced at the deciduo-placental interface (142, 379) and may generate vasoinhibins from decidual PRL. Indeed, amniotic fluid accumulates decidual PRL (46) and contains PRL-derived vasoinhibins, and cathepsin D from placental trophoblasts cleaves PRL to vasoinhibins (210). Likewise, the antiangiogenic vasodilator ANG-(1–7) and its generating enzyme, ACE2, are present in the decidua during early placentation when the ANG II/ANG-(1–7) ratio decreases at the implantation and interimplantation sites (395) and in various cell types of the placenta at term (574).
3. Mammary gland
The mammary gland undergoes dramatic changes in growth, differentiation, function, and involution during the reproductive cycle (see reviews in Refs. 240, 503). Pituitary gland hormones and ovarian steroids orchestrate these changes starting at puberty, when gonadotropins promote the ovarian release of estrogen and progesterone that stimulate the elongation and branching of the mammary gland ductal system. In addition, adrenocorticotropin hormone stimulates ductal growth in cycling females through glucocorticoid and GH release (see reviews in Refs. 250, 282). During pregnancy, progesterone, PRL, and PL promote the proliferation, differentiation, and maturation of the alveolar system responsible for milk secretion (240, 409), and while PRL appears essential for the initiation and maintenance of lactation in most species, milk production is controlled predominantly by GH in ruminants (44, 169). After weaning, involution returns the mammary gland to a virgin state, and both PRL and GH help regulate mammary gland survival by protecting it against epithelial cell apoptosis and ECM degradation (31, 168, 169).
The growth and function of the mammary gland depends on an increasing supply of oxygen, nutrients, and fluid and is therefore accompanied by the expansion of the mammary gland vasculature during pregnancy and enhanced vasodilation and vasopermeability during lactation (136, 352, 432), whereas this new vascular bed progressively disappears during the involution phase (432, 587). The importance of angiogenesis and specifically of VEGF in the mammary gland is illustrated by the fact that inactivation of the VEGF gene in the mammary gland epithelium impairs angiogenesis and blood vessel function during lactation and leads to reduced milk production and stunted growth of the offspring (472). The fact that hormones with effects on angiogenesis control the growth and function of the mammary gland makes it an attractive model for the study of the hormonal regulation of angiogenesis.
VEGF and VEGFR2 increase in the mammary gland during pregnancy and even more during lactation (432), when the influence of PRL appears to be greatest (44). PRL promotes the expression of VEGF in HC11 mouse mammary epithelial cells (207) and in immune cells associated with mammary gland involution (207, 344). Furthermore, both PRL (301, 520) and GH (450) are expressed in mammary gland epithelial cells and may regulate VEGF levels in an autocrine manner. Consistent with these findings, human MCF-7 mammary carcinoma cells overexpressing human GH produce higher levels of VEGF that promote angiogenesis in vivo and in vitro (67). The proangiogenic actions of these hormones may be counterbalanced by their conversion to vasoinhibins, since cathepsin D-mediated generation of vasoinhibins from PRL increases in the lactating mammary gland (91, 325), and the expression of various vasoinhibin-generating MMPs is upregulated during mammary gland development and involution (216). PRL, GH, PL, and vasoinhibins can also regulate mammary gland growth and function by controlling peripheral vascular resistance and blood flow to the gland. PRL can stimulate either vasodilation or vasoconstriction, and vasoinhibins have clear vasoconstrictive effects (see review in Ref. 96). These vasomotor actions also occur in response to human GH (572) and PL (221).
Likewise, members of the RAS and KKS can affect mammary gland function. BK regulates mammary gland blood flow (448, 637), and both BK and ANG II act as growth factors for normal mammary gland epithelial cells (214, 215). ANG II present in mammary gland is not necessarily derived from the circulation, as RAS components have been localized in normal mammary gland tissue, particularly in epithelial cells (538). Also, milk contains BK (613), and the isolated lactating bovine udder releases BK into the vascular perfusate (637). To our knowledge, the role of the RAS and KKS in normal mammary gland angiogenesis has not yet been investigated.
B. Nonreproductive Organs
With the exception of the female reproductive system, the vasculature of most healthy tissues is quiescent during adult life, reflecting the predominance of naturally occurring inhibitors able to counterbalance the effects of relatively abundant proangiogenic mediators. The study of this tight control of angiogenesis is particularly attractive in tissues like retina and cartilage, which are partially or totally devoid of blood vessels, respectively, and where damage to the mechanisms regulating angiogenesis contributes to the development of vasoproliferative retinopathies and arthritis.
In the normal adult retina, the vascularized compartment is confined to the inner organ, whereas the outer retina never becomes vascularized. Failure to inhibit blood vessel growth can result in reduced visual acuity and underlies retinopathy of prematurity, diabetic retinopathy, and age-related macular degeneration, the leading causes of blindness worldwide (642). Among the multiple regulators of retinal angiogenesis (122, 138) evidence suggests that vasoinhibins derived from PRL are crucial (see review in Ref. 95). PRL and vasoinhibins are present in ocular fluids and in the retina of rats (22, 464) and humans (140, 441) and may originate from systemic PRL (407) or from PRL synthesized locally, as PRL mRNA and protein are localized in cells throughout the retina (22, 464), including endothelial cells (410). Notably, the intravitreal injection of antibodies able to inactivate vasoinhibins promotes vessel growth in the retina, and the intraocular transfection of siRNA to block PRL expression stimulates retinal angiogenesis and vasodilation (22). Endogenous vasoinhibins may also participate in the control of vessel remodeling during development. Immunosequestering vasoinhibins in neonatal rats reduced the apoptosis-mediated regression of hyaloid vessels, a transiently existing intraocular system of blood vessels that nourishes eye tissues during the embryonic period (140).
An important aspect of the retinal vasculature in mice is that its development begins during the first week after birth, making the murine retina a valuable model in which to study the mechanisms governing the whole angiogenesis process under physiological conditions (569). Vascularization originates at the optic nerve in the superficial layer of the inner retina and radiates towards the periphery, using a mesh of migrating astrocytes as a template and proangiogenic factors released by ganglion cells, astrocytes, and endothelial cells themselves as a guide (193, 569). GH may be among the proangiogenic factors promoting vascular development in the retina, since GH and the GH receptor are expressed in the ganglion cell layer and inner nuclear layer of the newborn mouse retina at the time when retinal vascularization occurs, and GH receptor-deficient mice show a reduction in the width of the retina and altered levels of proteins involved in retinal vascularization, i.e., protein kinase C inhibitor 1, cyclophilin A, and Sam68-like mammalian protein-2 (41). Furthermore, patients with defects in the GH/IGF-I axis exhibit reduced retinal vascularization (238, 239). The GH affecting retinal vascular development may also be derived from the circulation, since GH levels in the vitreous correlate with those in the systemic circulation in neonatal rats (43, 369). In addition, the RAS may promote retinal angiogenesis in the retina during development, when AGT, prorenin, ANG II, and the AT1 and AT2 receptors are localized in cells and blood vessels of the ganglion cell layer of the inner retina (485). The hypertensive transgenic m(Ren-2)27 rat model, in which renin and ANG II are elevated in various tissues including the retina (375), shows a more extensive development of the peripheral retinal vasculature and a reduction in vascular density in the immature retina after ACE inhibition (485).
Members of RAS are also expressed in the adult retina of mammals, including humans and rats (295, 584, 605), and there is evidence for the local production of the antiangiogenic ANG-(1–7) metabolite in the human retina (490). Furthermore, some components of the tissue KKS are expressed in neuronal cells of the outer nuclear layer, inner nuclear layer, and ganglion cell layer of the adult retina of humans (336) and rats (540), but the possible effects of counterbalancing RAS and KKS upon physiological angiogenesis in the retina remain undefined.
Cartilage is resistant to vascular invasion except during endochondral bone formation (120, 228) or in degenerative joint diseases such as rheumatoid arthritis (293). Avascularity helps provide the elasticity, flexibility, and strength of cartilage and results from the action of a variety of locally produced antiangiogenic factors that overcome the effect of multiple proangiogenic mediators (228, 500). Vasoinhibins are among the antiangiogenic factors produced in cartilage, since PRL is a component of synovial fluid (411), chondrocytes are enriched in the MMPs that convert PRL to vasoinhibins, and PRL mRNA, PRL, and vasoinhibins have been detected in articular chondrocytes (339). Articular chondrocytes also express PRL receptors, and PRL promotes chondrocyte survival (638), suggesting that both PRL and vasoinhibins act to maintain the functional integrity of cartilage.
During endochondral ossification, hypertrophic chondrocytes of the growth plate switch their phenotype from antiangiogenic to proangiogenic, producing factors including VEGF that attract blood vessels (120). The invading blood vessels bring progenitor mesenchymal cells that will later differentiate into osteoblasts and chondroclasts/osteoclasts to remodel the newly formed cartilage into bone (120). Endochondral ossification leads to longitudinal bone formation, a process governed by an intricate system of endocrine signals, including the GH/IGF-I axis (see review in Ref. 402). GH promotes longitudinal bone growth by IGF-I-independent and -dependent effects on growth plate chondrocyte proliferation and hypertrophy, respectively (595). It remains to be determined whether proangiogenic effects of the GH/IGF-I axis contribute to their actions on endochondral ossification.
V. HORMONAL CONTRIBUTION TO ANGIOGENESIS-DEPENDENT DISEASES
As previously noted, tight regulation of angiogenesis is required to prevent either excessive vascular growth or aberrant vessel regression, each of which can lead to various diseases. Consequently, the angiogenic and antiangiogenic actions of members of the GH/PRL/PL family, the RAS, and the KKS have been linked to both the etiology and treatment of angiogenesis-related pathologies.
Preeclampsia is defined as the onset of hypertension and proteinuria after 20 wk of gestation that, if left untreated, can progress to eclampsia, a state of generalized seizures that may harm or kill the mother or the unborn child. This disease affects 5% of all pregnancies and is a major cause of maternal, fetal, and neonatal morbidity and mortality (502). Although the etiology of preeclampsia remains undefined, it is clear that the disease depends on the placenta, as all symptoms disappear after its delivery.
Studies carried out during the last decade have shown that antiangiogenic factors produced by the placenta are central to the pathophysiology of preeclampsia (for reviews, see Refs. 128, 199, 456). The syndrome is thought to arise at an early stage of pregnancy due to placental ischemia/hypoxia produced by defective trophoblast remodeling of the uterine spiral arteries. Hypoxia, in association with oxidative stress and other placental abnormalities (259, 456), stimulates the release of antiangiogenic molecules such as sVEGFR1 (247, 316, 355), soluble endoglin (sEng) (315, 318, 580), and possibly autoantibodies against AT1 receptors (AT1-AA) (259, 594). sVEGFR1 decreases free VEGF and placental growth factor levels in preeclampsia, leading to endothelial cell dysfunction and inhibition of vasodilation (331, 355). Eng is part of the TGF-β receptor complex, and sEng is antiangiogenic and inhibits eNOS-mediated vasodilation (580). Furthermore, the adenoviral expression of sVEGFR1 in pregnant rats induces the clinical signs of preeclampsia, i.e., hypertension, proteinuria, and glomerular endotheliosis (355), and the coexpression of both sVEGFR and sEng exacerbates endothelial cell dysfunction leading to severe preeclampsia, including the HELLP (hemolysis, elevated liver enzymes, low platelets) syndrome and restriction of fetal growth (580). In addition, AT1-AA with agonistic properties (57, 624) increases in the circulation of women with preeclampsia (594), and ANG II can stimulate sVEGFR1 expression in trophoblasts via AT1 receptors (647).
The influence of RAS on the pathophysiology of preeclampsia is indicated by the development of a preeclampsia-like syndrome in transgenic mice overexpressing placental renin and maternal AGT (541) and involves multiple mechanisms (see reviews in Refs. 259, 493). Although RAS components exist in the maternal and fetal placenta (108, 317) and at high levels in the circulation (33, 241), vascular responsiveness to ANG II is reduced during pregnancy (188). In contrast, components of RAS are lower in plasma (66, 305), and vascular responsiveness to ANG II is enhanced in patients with preeclampsia (417). Increased sensitivity to ANG II can contribute to preeclamptic characteristics, including abnormal trophoblast invasion and spiral artery remodeling, reduced uteroplacental blood flow, systemic vasoconstriction, and hypertension (259). Increased ANG II sensitivity may involve elevated AT1 receptor expression, which is reported to occur in the decidua of preeclamptic women (241), but, more importantly, heterodimers between AT1 receptors and BK B2 receptors are more abundant in blood vessels during preeclampsia (3). AT1-B2 heterodimers induce increased ANG II-mediated signaling in blood vessels (3) and are resistant to inactivation by oxidative stress (2, 3, 24). On the other hand, an altered production of the antiangiogenic, vasodilating ANG-(1–7) metabolite may also counterbalance the actions of ANG II in preeclampsia, since its levels increase and decrease in the circulation during normal pregnancy and preeclampsia, respectively (64, 578). Moreover, the increased vascular peripheral resistance in preeclampsia may reflect reduced responsiveness to BK, since BK-induced vasodilation of small peripheral arteries is enhanced in normal pregnancy and impaired in preeclampsia (288). However, the influences of ANG-(1–7) and BK in preeclampsia remain poorly understood.
A role for PRL in the pathogenesis of preeclampsia was suggested 30 years ago based on its osmoregulatory and hypertensive properties (248), and interest in this hypothesis was renewed with the discovery of the antiangiogenic and vasoconstrictive effects of vasoinhibins (428). Although the maternal, fetal, or amniotic fluid levels of PRL do not change in preeclampsia (333, 453), a recent study showed that urinary PRL increases in preeclamptic women in relation to the severity of the disease (310). The association between PRL levels and adverse outcomes in preeclampsia may reflect its conversion to vasoinhibins. Vasoinhibins are enhanced in the amniotic fluid, serum, and urine of preeclamptic women and may derive from the cathepsin D-mediated cleavage of PRL in the preeclamptic placenta (210, 310). Notably, the concentrations of PRL and vasoinhibins in the amniotic fluid follow an inverse correlation with birth weight in preeclampsia (210), indicating the association of these proteins with the clinical manifestation of the disease. Moreover, although much needs to be learned in this regard, there is compelling evidence that overproduction of vasoinhibins can lead to postpartum cardiomyopathy, another antiangiogenesis-dependent disease in reproduction similar to preeclampsia in that oxidative stress is a key factor for its etiology (243). In postpartum cardiomyopathy, cardiomyocyte-specific deletion of STAT-3 promotes oxidative stress, thereby enhancing cathepsin D-mediated generation of vasoinhibins, which in turn interfere with the coronary microvasculature growth and function required for adequate performance of the maternal heart during pregnancy and lactation (243).
B. Diabetic Retinopathy
Diabetic retinopathy is the leading cause of blindness in working-age individuals throughout the world. After 20 years, more than 60% of patients with type 2 diabetes and nearly all patients with type 1 diabetes develop retinopathy (171). The major risk factor is chronic hyperglycemia, which causes the apoptosis of pericytes and endothelial cells, the thickening of capillary basement membranes, and enhanced vasopermeability. Hyperpermeability leads to abnormal retinal hemodynamics and to the accumulation of extracellular fluid and hard exudates that impair vision when the macula is affected (290). Over time, intraretinal hemorrhages and capillary occlusion create areas of ischemia; the resulting retinal hypoxia leads to the production of local proangiogenic factors. In the more advanced stages, the new blood vessels invade and bleed into the vitreous, producing a fibrovascular tissue that may result in tractional retinal detachment and blindness. The main therapy for diabetic retinopathy is laser photocoagulation, which limits vascular leakage in the retina and blocks the induction of proangiogenic factors by ablating retinal ischemic areas (628). Although laser photocoagulation is effective in preventing visual loss in many cases, it rarely improves visual acuity, may damage peripheral and night vision, and often does not quell progression of the disease (56). Thus new pharmacological therapies are being developed to target microvascular abnormalities, including anti-VEGF agents, ACE blockers, and GH inhibitors.
VEGF is a major factor enhancing vasopermeability and inducing angiogenesis in diabetic retinopathy (68). It is upregulated in the retina of animals with experimental retinopathies (363, 437), and its levels increase in the ocular fluids of patients with diabetic retinopathy (10, 507). Elevation of intraocular VEGF results in vasodilation, retinal hemorrhages, vascular leakage, and neovascularization in nondiabetic animals (100, 192, 368, 413, 558), and therapies based on the inhibition of VEGF expression and action have shown promising results as effective and safe options for the treatment of vasoproliferative retinopathies (149, 506, 628). Although anti-VEGF therapies still need to be evaluated in phase III clinical trials, their use has been authorized for age-related macular degeneration, and they are currently employed by many ophthalmologists to treat advanced diabetic retinopathy and diabetic macular edema (506).
Evidence for the role of RAS in the pathogenesis of diabetic retinopathy has been the subject of comprehensive reviews (187, 608, 609) and stems from observations showing that hypertension is a risk factor for the development and progression of diabetic retinopathy (287), that chronic hyperglycemia activates RAS (19), that RAS components are elevated in the plasma and ocular fluids of patients with diabetic retinopathy (143, 187, 608) and, importantly, that ACE inhibitors interfere with experimental retinopathy and reduce the onset and progression of the disease in humans (608). ANG II promotes diabetic retinopathy by stimulating systemic hypertension and retinal hyperperfusion and pressure, leading to shear stress-mediated release of VEGF from retinal vessels (532). However, the recent observation that an AT1 antagonist, but not antihypertensive therapy, ameliorates vascular pathology in the retina of certain diabetic rats implicates actions of RAS that are independent of high blood pressure (611). Indeed, ANG II directly stimulates the expression of VEGF and VEGFR2 in retinal cells (420, 643) and potentiates VEGF-induced proliferation of retinal capillary vessels in vitro (421). ACE inhibition reduces retinal VEGF expression and hyperpermeability (143, 200, 643) and restores retinal blood flow (245) in diabetic rats. ACE inhibition also prevents retinal neovascularization and decreases VEGF, VEGFR2, and angiopoietin-2 expression in murine ischemia-driven retinopathy (327, 374, 388, 484) in diabetic, transgenic rats overexpressing renin and AGT in the eye (375), and in the vitreous of patients with diabetic retinopathy (244).
In spite of the above experimental evidence, blockage of RAS has shown little or no effect on retinal blood vessel abnormalities in clinical trials. Improvement of diabetic retinopathy was reported after RAS blockage in patients with normotensive type 1 diabetes (85, 86, 306), but no beneficial effect was reported in patients with hypertensive type 1 diabetes (151), nor in most studies of patients with type 2 diabetes (609). The reasons for this are unclear and may involve interaction with other members of RAS and the KKS (127, 436). For example, activation of ACE2 can counterbalance the effects of ACE and ANG II by causing ANG II degradation and the production of ANG-(1–7) (Fig. 3). ACE2 expression is reduced in the diabetic retina (557), suggesting that the deleterious effect of ANG II may be exacerbated by local ACE2 deficiency. On the other hand, ACE catalyzes not only ANG II formation, but also the degradation of the vasodilator and proangiogenic peptide, BK. Recent evidence shows that BK increases vasopermeability in the healthy retina, that B1 receptors are upregulated in the retina of diabetic rats in response to oxidative stress, and that the blockage of both B1 and B2 receptors prevents the breakdown of the blood-retinal barrier associated with experimental diabetes (4). Furthermore, treatment with ACE inhibitors blocks oxidative stress and the induction of B1 receptors in diabetic rats (114), suggesting that these actions prevent the deleterious effects of BK accumulated in response to ACE inhibitors. Importantly, members of the KKS (prekallikrein, kallikrein, FXII, FXIIa, and HK) are present in the vitreous of individuals with advanced diabetic retinopathy and are activated in response to and mediate hemorrhage-induced retinal edema (189).
The contribution of the GH/IGF-I axis to the pathophysiology of diabetic retinopathy (180, 612) was described nearly four decades ago based on the elevated circulating GH levels found in diabetic subjects (275, 445, 497), the halted progression of diabetic retinopathy after anterior pituitary ablation (335) or radiation (496), and the reduced incidence of the disease in GH-deficient diabetic patients (14). Moreover, evidence that IGF-I reverses protection by a GH antagonist against ischemia-induced retinopathy (510), promotes VEGF-mediated retinal angiogenesis (443, 511), and induces most of the alterations seen in diabetic retinopathy (475) substantiates the causative role of the GH/IGF-I axis in the development of the disease.
Although IGF-I circulating levels increase as a function of the progression of diabetic retinopathy in type 1 diabetic women during pregnancy (307), most studies in nonpregnant, diabetic patients have failed to show a relationship between circulating hormones and diabetic retinopathy. Various studies have demonstrated an increase (135, 268), decrease (157), or lack of change (181, 596) in circulating IGF-I in association with the presence or progression of the disease. Clinical trials with the GH antagonist pegvisomant yielded negative results (222), and the use of somatostatin analogs to block anterior pituitary GH secretion and to promote local somatostatin antiangiogenic actions on endothelial cells generated varying outcomes (see review in Refs. 213, 424). The reasons for these discrepancies are unclear and are likely influenced by small patient cohorts confounded not only by systemic factors such as glucose control, aggressive insulin treatment, and differential renal function, but also by ocular factors that modify the systemic incorporation or the local synthesis of GH, IGF-I, and IGF-I binding proteins (180, 612).
Little is known regarding the involvement of PRL in diabetic retinopathy (95). PRL may protect against diabetes, as suggested by the facts that PRL and PL promote the function, proliferation, and survival of β-cells (166, 177, 183), and circulating PRL levels are often decreased in poorly controlled diabetic patients (260) and rats (60). However, studies measuring circulating PRL in association with diabetic retinopathy have reported increased (373), decreased (253), or normal (81) levels. These contradictory observations may be due in part to the intraocular generation of vasoinhibins. Systemic PRL increases and gets processed to vasoinhibins in ocular fluids and in retrolental fibrovascular membranes of patients with retinopathy of prematurity (140), a neovascular eye disease caused by elevated oxygen used to improve the survival of premature neonates. Notably, vasoinhibins prevent retinal angiogenesis in ischemia-induced retinopathy (425) and inhibit the retinal vasopermeability associated with diabetic retinopathy (192). Nevertheless, studies addressing the endogenous levels of vasoinhibins and their proteolytic generation from PRL, GH, or PL in diabetic retinopathy are still needed.
C. Rheumatoid Arthritis
Rheumatoid arthritis is a chronic inflammatory autoimmune disorder of unknown etiology that targets the synovial membrane, cartilage, and bone and affects 1% of the world population (165). Autoimmunity followed by the articular infiltration of leukocytes and the hyperplasia of synovial cells lead to the development of an invasive, inflammatory front, or “pannus” that destroys the adjacent cartilage and bone. The hypoxic environment of the arthritic synovium and the action of a variety of cytokines, chemokines, and regulatory factors promote the expression of proangiogenic substances, including VEGF (113, 534). Subsequent angiogenesis facilitates the progression of the disease by enabling the continued accumulation of immune cells, expansion of the inflamed tissue, and swelling of the joints (32, 113). Indeed, blockage of VEGF action (6, 332, 365, 513) ameliorates experimental arthritis, and various antirheumatic drugs have antiangiogenic properties (576). However, no clinical approval has been obtained to use anti-VEGF therapies or other antiangiogenic therapies in patients with rheumatoid arthritis.
Epidemiological, experimental, and clinical data have linked GH and PRL to the pathophysiology of rheumatoid arthritis (reviewed in Refs. 347, 393). Interest in GH is raised by the observation that growth retardation is a major problem in juvenile rheumatoid arthritis (52), whereas emphasis on PRL derives from rheumatoid arthritis being three times more frequent in women than in men and the disease becoming exacerbated in the postpartum period in association with breast-feeding (229). Notably, both GH and PRL have immunoenhancing properties (357, 635) and restore experimental arthritis in hypophysectomized rats (50, 392); treatments with somatostatin and bromocriptine, which suppress pituitary secretion of GH and PRL, respectively, are beneficial in experimental (353, 354) and clinical (163, 164) arthritis. Levels of GH and PRL have been shown to increase in the serum (492, 562, 650) and synovial fluid (131, 387) of patients with rheumatoid arthritis. However, this issue is controversial (452, 473, 474), and no experimental evidence has demonstrated the proangiogenic actions of GH and PRL in the development of rheumatoid arthritis. Of note, the finding that MMPs capable of generating vasoinhibins are upregulated in the arthritic joint (339, 539) and that some of them are activated by locally produced PRL (387) raises the possibility that vasoinhibins generated in rheumatoid arthritis could help counterbalance the immunoenhancing and proangiogenic properties of the full-length hormones.
The contribution of RAS to the pathology of rheumatoid arthritis is suggested by the observations that ANG II has proinflammatory effects (476), that ACE, ANG II, and AT1 receptors are upregulated in synovial tissue from experimental arthritis (446) or from patients with rheumatoid arthritis (589, 590), and that targeting the angiotensin pathway with ACE blockers and AT1 inhibitors attenuates experimental arthritis (121, 479) and human inflammatory synovitis (446). Blockers of ACE activity may exacerbate inflammation by preventing the breakdown of BK. All components of the KKS have been detected in the synovial fluid of patients with rheumatoid arthritis (see review in Ref. 77) and are elevated in the peptidoglycan polysaccharide-induced rat model of arthritis and systemic inflammation (154). In this model, a BK inhibitor (105) or blockage of BK generation by an anti-HK monoclonal antibody (154) prevents joint swelling and inflammation. BK inhibitors also suppress synovitis and joint erosion in collagen-induced arthritis in mice (182). The B2 receptor is present in the normal synovia, and the B1 receptor is upregulated in arthritic tissue (40, 489); the B2 receptor has been proposed to promote angiogenesis in the early stages of synovitis, whereas the B1 receptor mediates endothelial cell proliferation in the established disease (489).
As in normal tissues, tumors depend on the blood supply for oxygen, nutrients, and waste removal so that in the absence of angiogenesis, tumor growth is limited to 1–2 mm in diameter (639). A key mechanism for tumor angiogenesis is local hypoxia resulting from the persistent and uncontrolled proliferation of tumor cells (231). Hypoxia triggers the expression of proangiogenic factors such as VEGF, placental growth factor, and bFGF that, when combined with oncogene-driven expression of proangiogenic peptides and inhibition of antiangiogenic agents, tilt the angiogenic balance in favor of new vessel growth (170, 498).
VEGF is overexpressed in 60% of all human cancers and is the target of antiangiogenic drugs currently used for cancer therapy that include a monoclonal anti-VEGF antibody adapted for human use (bevacizumab) and tyrosine kinase inhibitors selective for VEGF receptors (170). The survival benefits of anti-VEGF therapy are small, transitory, and accompanied by some toxic side effects (145, 283, 328). The benefits may be limited by the fact that the tumor microenvironment contains a variety of angiogenesis regulators besides VEGF (170). Insights into the role of hormones may help clarify the mechanisms mediating tumor angiogenesis and contribute to improving antiangiogenic therapies so that they continue to be effective when resistance to VEGF inhibition develops.
The contribution of pituitary gland hormones to mammary gland tumorigenesis was revealed more than 50 years ago when breast cancer was treated by hypophysectomy (334, 454). The regression of breast cancers resistant to anti-estrogen therapy following hypophysectomy (430) and the fact that PRL restored vulnerability to mammary tumorigenesis in hypophysectomized rats (21) pointed to PRL as the putative pituitary hormone essential for mammary tumor development. The participation of PRL in mammary cancer has been established in rodents, where hyperprolactinemia or PRL treatment increases the number of spontaneous mammary tumors and susceptibility to mammary carcinogens, while decreasing systemic PRL inhibits mammary tumor growth (see reviews, in Refs. 232, 583, 600). Furthermore, immunoneutralization of PRL inhibits the development of carcinogen-induced mammary tumors (362), transgenic mice overexpressing PRL develop mammary carcinomas (470, 601), and lack of PRL receptors reduces the size of mammary neoplastic growths (408).
PRL was initially considered irrelevant due to contrasting data provided by multiple clinical studies (reviewed in Ref. 101). Such variability was recently overcome by large, well-controlled studies showing that higher plasma PRL levels significantly increase the relative risk of developing breast cancer in premenopausal and postmenopausal women (230, 565, 566). In addition, epidemiological studies support the association of both dopamine antagonists and hyperprolactinemia with increased breast cancer risk (reviewed in Ref. 232). Of importance, however, neoplastic breast tissue synthesizes PRL, which is unaffected by dopamine, and it is recognized that PRL acts partly as an autocrine/paracrine promoter of mammary tumor growth (for review, see Refs. 45, 101, 583). Up to 98% of breast tumors express PRL receptors, and receptor levels are higher in neoplastic tissue than in adjacent normal tissue (201, 561). PRL stimulates the growth and survival of breast cancer cells (101, 433), and it promotes the expression of VEGF in mammary gland cell lines (207); thus increased angiogenesis may contribute to the stimulatory action of PRL on breast cancer. Mammary carcinomas in organ culture have diminished ability to generate vasoinhibins from PRL (34), and hypoxic conditions, mimicking the tumor microenvironment, are associated with reduced secretion of cathepsin D by a cultured pituitary adenoma cell line (112). These findings are in contrast to the reported greater ability of tumor cells to express and secrete cathepsin D and MMPs (423, 467), but imply that levels of vasoinhibins could be reduced in the microenvironment of tumors, thus creating a more favorable angiogenic condition for tumor progression. Notably, colon (49) and prostate (285) cancer cells overexpressing vasoinhibins generate smaller and less vascularized tumors in mice.
In addition to PRL, the GH/IGF-I axis has been shown in recent studies to play a key role in mammary cancer, particularly in animal models (see reviews in Refs. 434, 598, 627). GH restores vulnerability to mammary carcinogens in hypophysectomized rats (634) and in GH-deficient rats and mice (499, 533, 552). Disruption of the GH receptor gene inhibits oncogene-driven mammary tumorigenesis (644), and transgenic mice overexpressing a GH antagonist (442) and mice with a liver-specific IGF-I gene deletion (618) exhibit low incidence of chemically induced mammary tumors. Along this line, clinical studies have reported an association between height at various stages of development and breast cancer risk (225). High levels of both GH (148) and IGF-I (435) have been reported in the serum of breast cancer patients, and a correlation has been noted between elevated circulating IGF-I and increased risk of breast cancer development in women (reviewed in Ref. 585). GH may act as an oncogene, since its expression in the normal human mammary gland increases with the acquisition of proliferative lesions, and the experimental overexpression of GH transforms a human, nontumorigenic mammary epithelial cell line into a tumorigenic, invasive phenotype (382, 434, 648).
In support of angiogenesis being among the tumorigenic effects of GH, receptors for GH are found in endothelial cells of newly formed tumor capillaries (550), and the overexpression of GH in human mammary carcinoma cells stimulates endothelial cell migration and tube formation in vivo dependent of VEGF (67). Also, autocrine GH downregulates the production of the antiangiogenic factor thrombospondin (626). The direct actions of GH in promoting tumor growth may involve cross-talk with the PRL receptor, which binds human GH. By overexpressing receptor-specific ligands it was shown that the PRL receptor, but not the GH receptor, mediates the high incidence of chemically induced mammary tumors in mice (601).
In contrast to PRL and GH, for which little is known in malignancies other than breast cancer (for reviews see Refs. 37, 45, 232, 627), substantial information relates RAS members to a wide variety of cancers. The extensive use of ACE inhibitors and AT1-receptor blockers to treat hypertension and congestive heart failure allowed the accumulation of clinical data that, nonetheless, is inconclusive (see reviews in Refs. 7, 323). While some studies showed that ACE inhibitors or AT1 receptor blockers reduce the incidence of cancer of the breast, female reproductive tract, lung (314), esophagus (508), and prostate (469), other studies have failed to establish such an association (179, 324, 358). Variability may relate to differences in population profiles, type of inhibitors, pharmacological parameters, and follow-up times. Also, the type of neoplasia appears relevant since the levels and activity of RAS members increase in concert with the progression of some cancers but not with others (7, 133, 153, 258, 329, 466). A naturally occurring deletion in the ACE gene leading to its elevated expression correlates with increased susceptibility to prostate (630) and breast cancer (575), but not to colorectal and lung cancer.
Experimental studies have provided stronger support for the role of RAS blockers as anticancer agents. ACE inhibitors and/or AT1 receptor antagonists inhibit tumor growth and angiogenesis in murine models of fibrosarcoma (582), renal carcinoma (242, 367), glioblastoma (463), hepatocellular carcinoma (631), head and neck squamous cell carcinoma (629), Lewis lung cancer (185), colorectal cancer metastasis to the liver (394), melanoma (125, 144), prostate cancer (298, 570), ovarian cancer (530), and gastric tumors (252). Blockage of RAS is frequently associated with reduced levels of VEGF (144, 298, 582), and both VEGF expression and action can be promoted by ANG II-induced activation of AT1 receptors (236). Studies comparing the growth and neovascularization of syngeneic tumors growing in wild-type and AT1 receptor null-mice have shown that AT1 receptors in host cells contribute to tumor growth and angiogenesis via VEGF synthesis (184, 256). In addition, long-term blockage of AT1 receptors leaves the AT2 receptor fully activated (606) and, because there is evidence that AT2 receptors mediate ANG II-induced inhibition of angiogenesis (47), the selective activation of only AT2 receptors may contribute to the antiangiogenic effect of AT1 blockers. However, AT2 receptors may also be proangiogenic (465), and an alternative pathway by which AT1 blockers may prevent angiogenesis is by increasing the concentration of ANG II and thus its conversion to antiangiogenic ANG-(1–7) (71, 294, 377). ACE inhibitors can also increase circulating ANG-(1–7) by enhancing ANG I levels and preventing ACE-mediated degradation of ANG-(1–7) (71, 83). On the other hand, inhibition of renin, leading to the accumulation of antiangiogenic AGT, may help inhibit tumor angiogenesis. Adenovirus-mediated gene transfer of AGT inhibits the growth, neovascularization, and metastasis of mammary carcinomas and melanomas in mice (61). Therefore, the outcome of RAS effects on tumor angiogenesis depends on the balance between its various metabolites with opposite effects on blood vessel growth. It is noteworthy that a common side effect of anti-VEGF cancer therapies is hypertension (328), and ACE inhibitors are frequently used in clinical trials of anti-VEGF therapies with no reference to whether these treatments also influence tumor angiogenesis (7).
In addition to interfering with the RAS system, ACE inhibitors can affect tumor growth and angiogenesis by inhibiting ACE-mediated BK degradation. Indeed, sarcoma neovascularization and growth is enhanced by the ACE inhibitor captopril, and these effects are attenuated by blocking the KKS with plasma kallikrein inhibitors or B2-receptor antagonists (261). BK is increased in plasma and ascitic fluid of some cancer patients (342, 343). Experimental data support the contribution of the KKS to tumor angiogenesis. In mice, BK enhances vascular permeability and VEGF expression in solid and ascitic tumors (262, 617), and immunoneutralization of HK to block its cleavage to BK inhibits the growth and neovascularization of human colon carcinoma cells (514) and murine multiple myeloma cells (480). Also, inoculation of Walker 256 carcinoma cells into kininogen-deficient rats that cannot generate BK results in smaller and less-vascularized tumors with a reduced VEGF content, characteristics that are mimicked by the treatment of carcinoma-bearing, normal rats with a B2-receptor antagonist (255). B2 receptors are highly upregulated in human and rodent cancer tissues (54, 616), and tumor angiogenesis and growth are markedly reduced in B2 receptor-knockout mice bearing sarcomas compared with their wild-type littermates (255). B1 receptors are found only in malignant tumors (548), and evidence of their contribution to cancer comes from studies in mice showing that both B1 and B2 receptors are required for the growth of androgen-insensitive prostate cancer cells (39, 522) and that a BK antagonist able to block both B1 and B2 receptors inhibits lung cancer more potently than does a VEGF-receptor inhibitor (82).
Blockage of BK synthesis may also prevent generation of the other product of HK cleavage, the antiangiogenic protein HKa. Experimental work shows that the antiangiogenic domain 5 of HKa suppresses lung metastasis experimentally induced by a malignant melanoma cell line (279). There is limited experience with BK antagonists in the clinic (reviewed in Ref. 521) and none in relation to cancer. Both peptide and nonpeptide antagonists of BK are being developed that are more efficient and less toxic than many of the current anticancer drugs (523).
VI. CONCLUSIONS AND FUTURE DIRECTIONS
Members of the GH/PRL/PL family, the RAS, and the KKS regulate a wide spectrum of biological effects that depend on blood vessel number and function, including organ growth and involution, vascular permeability, and inflammation. In recent years it has become evident that these hormonal systems regulate angiogenesis by exerting both stimulatory and inhibitory influences, but the mechanisms of these actions are not well understood. The complexity of hormonal involvement is evidenced by the fact that regulation of angiogenesis depends on diverse ligands, their receptors, and their multiple signaling pathways. Also, the overlapping functions of members within and among these hormonal families influence the outcome of the angiogenic response.
Proteolytic cleavage is a mechanism shared by the PRL/GH/PL family, the RAS, and the KKS for the release of both positive and negative regulators of angiogenesis. Such cleavage can efficiently balance blood vessel growth and regression under physiological conditions, particularly in the female reproductive system, where members of these hormonal systems influence the growth and involution of the various organ structures. In addition, high circulating levels of the antiangiogenic hormonal derivatives help to maintain the quiescent state of blood vessels in adult life, while their downregulation in favor of proangiogenic metabolites contributes to angiogenesis-dependent diseases (Fig. 5).
Substantial evidence supports the action of PRL/GH/PL/vasoinhibins, RAS, and KKS in controlling angiogenesis in the reproductive organs and retina, although their putative effects in cartilage require further study. In addition, various experimental approaches including the use of powerful genetic models have highlighted the contribution of RAS in preeclampsia, of vasoinhibins in postpartum cardiomyopathy, of GH/IGF-I and RAS in diabetic retinopathy, and of PRL/GH, RAS, and KKS in cancer. The use of selective enzyme inhibitors and receptor blockers for RAS, dysfunctional mutations of the genes along the GH/IGF-I axis, and treatment with agents that target this axis at various levels have provided clinical data supporting the role of these hormones in the control of angiogenesis-related diseases.
Nonetheless, insufficient information concerning the regulation of angiogenesis by these hormonal families is available, few physiological and disease conditions have been examined in depth, many exceptions have been reported, and contrasting data are frequently encountered due, at least in part, to differences in animal models, tissue types, disease states, pharmacological parameters, and follow-up times. Tissue specificity is influenced by the relative proportion of stromal cells and immune cells releasing proangiogenic substances in response to hormone treatment. These effects also depend on the relative contribution of circulating versus locally produced hormones, the hormonal clearance rate, and the production and activity of the converting proteases. Data are surprisingly limited on the contribution of the endogenously occurring antiangiogenic moieties, i.e., vasoinhibins, AGT, des[ANG I]AGT, ANG-(1–7), and HKa, and on the presence and regulation of the converting proteases that determine their levels.
The study of the PRL/GH/PL family, the RAS, and the KKS opens numerous avenues for the pharmacological treatment of angiogenesis-related disorders. Drugs with minor side effects already available to modify the levels and effects of these hormones may be used to increase the levels of endogenous angiogenesis inhibitors or to block the action of proangiogenic mediators. So far, the complexity of these systems has prevented definitive conclusions regarding their pharmacological manipulation in human disease but has raised multiple challenges for the future. In view of the opposing effects exerted by the intact hormones and their proteolytically derived fragments, future research should focus on elucidating the mechanisms that regulate the expression and activity of the converting proteases and the selective expression of the receptors for each hormonal isoform together with their specific signaling pathways. Major research goals include the identification of functional domains affecting half-life, binding affinity, and biological potencies of the bioactive proteins to develop more specific and potent agonists and antagonists. Moreover, in-depth studies are needed to address the consequences of depleting or overproducing the various hormonal metabolites and their overlapping actions.
This review brings into focus the actions of the GH/PRL/PL family, the RAS, and the KKS on angiogenesis. Clearly, these hormonal systems are important in regulating the number and function of blood vessels, and numerous avenues of pharmacological intervention can be anticipated based on this research.
NOTE ADDED IN PROOF
During the review process of the manuscript, the work showing that cathepsin D is a primary protease for the generation of adenohypophysial vasoinhibins that was cited as unpublished observations (p. 1183) was accepted for publication (115a). Also, two seminal papers were published showing that VEGF inhibitors, while reducing primary tumor growth, promoted tumor invasiveness and metastasis (142a, 423a). These findings have raised significant concerns regarding the use of antiangiogenic therapies for cancer patients and highlight the need for a better understanding of local and systemic influences in the tumor and the blood vessel microenvironments that are altered by proangiogenic factor removal.
This work was supported by National Council of Science and Technology of Mexico Grants 87015, 44387, and 49292 and by UNAM Grant 200509.
We thank Guadalupe Calderón for artistic contributions, Gabriel Nava and Fernando López-Barrera for technical assistance, and Dorothy D. Pless for critically editing the manuscript.
Address for reprint requests and other correspondence: C. Clapp, Instituto de Neurobiología, Universidad Nacional Autónoma de México (UNAM), Campus UNAM-Juriquilla, Queretaro, Qro. 76230, Mexico (e-mail:).
- Copyright © 2009 the American Physiological Society